MoCa: Modality-aware Continual Pre-training Makes Better Bidirectional Multimodal Embeddings
Paper
• 2506.23115 • Published
• 36
text stringlengths 92 468 | images listlengths 1 1 |
|---|---|
<|vision_start|><|image_pad|><|vision_end|>
Represent the given image for classification
plane, carpenter's plane, woodworking plane | [
[
255,
216,
255,
224,
0,
16,
74,
70,
73,
70,
0,
1,
1,
0,
0,
1,
0,
1,
0,
0,
255,
219,
0,
67,
0,
8,
6,
6,
7,
6,
5,
8,
7,
7,
7,
9,
9,
8,
10,
12,
20,
13... |
<|vision_start|><|image_pad|><|vision_end|>
Represent the given image for classification
comic book | [
[
255,
216,
255,
224,
0,
16,
74,
70,
73,
70,
0,
1,
1,
0,
0,
1,
0,
1,
0,
0,
255,
219,
0,
67,
0,
8,
6,
6,
7,
6,
5,
8,
7,
7,
7,
9,
9,
8,
10,
12,
20,
13... |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n junco, (...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n zucchin(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n stingra(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n scuba d(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n miniski(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n grand p(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n paralle(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD//gAmRmlsZSB3cml0dGVuIGJ5IEFkb2JlIFBob3Rvc2hvcKggNC4w/9sAQwAIBgYHBgUIBw(...TRUNCATED) |
"<|vision_start|><|image_pad|><|vision_end|>\nRepresent the given image for classification\n syringe(...TRUNCATED) | ["/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAgGBgcGBQgHBwcJCQgKDBQNDAsLDBkSEw8UHRofHh0aHBwgJC4nICIsIxwcKDcpLD(...TRUNCATED) |
End of preview. Expand
in Data Studio
MMEB train split used in MoCa Continual Pre-training
🏠 Homepage | 💻 Code | 🤖 MoCa-Qwen25VL-7B | 🤖 MoCa-Qwen25VL-3B | 📚 Datasets | 📄 Paper
Introduction
This is a interleaved multimodal pre-training dataset used in the modality-aware continual pre-training of MoCa models. It is adapted from the train split of MMEB by concatenating queries and positive documents.
The dataset consists of interleaved multimodal examples. text is a string containing text while images are image binaries that can be loaded with the following code snippet:
import PIL.Image
from io import BytesIO
image_bytes = example['images'][0]
image = PIL.Image.open(BytesIO(image_bytes))
Citation
MoCa
@article{chen2025moca,
title={MoCa: Modality-aware Continual Pre-training Makes Better Bidirectional Multimodal Embeddings},
author={Chen, Haonan and Liu, Hong and Luo, Yuping and Wang, Liang and Yang, Nan and Wei, Furu and Dou, Zhicheng},
journal={arXiv preprint arXiv:2506.23115},
year={2025}
}
MMEB
@article{jiang2024vlm2vec,
title={VLM2Vec: Training Vision-Language Models for Massive Multimodal Embedding Tasks},
author={Jiang, Ziyan and Meng, Rui and Yang, Xinyi and Yavuz, Semih and Zhou, Yingbo and Chen, Wenhu},
journal={arXiv preprint arXiv:2410.05160},
year={2024}
}
- Downloads last month
- 6
Size of downloaded dataset files:
63.9 GB
Size of the auto-converted Parquet files:
63.9 GB
Number of rows:
1,448,799